Borate as a phosphate ester mimic in aldolase-catalyzed reactions: Practical synthesis of L-fructose and L-iminocyclitols

Article abstract of DOI:10.1002/adsc.200600356

Dihydroxyacetone phosphate (DHAP)-dependent aldolases have been widely used for the organic synthesis of unnatural sugars or derivatives. The practicality of using DHAP-dependent aldolases is limited by their strict substrate specificity and the high cost and instability of DHAP. Here we report that the DHAP-dependent aldolase L-rhamnulose 1-phosphate aldolase (RhaD) accepts dihydroxyacetone (DHA) as a donor substrate in the presence of borate buffer, presumably by reversible in situ formation of DHA borate ester. The reaction appears to be irreversible, with the products thermodynamically trapped as borate complexes. We have applied this discovery to develop a practical one-step synthesis of the non-caloric sweetener L-fructose. L-Fructose was synthesized from racemic glyceraldehyde and DHA in the presence of RhaD and borate in 92% yield on a gram scale. We also synthesized a series of L-iminocyclitols, which are potential glycosidase inhibitors, in only two steps.

Full text of DOI:10.1002/adsc.200600356

COMMUNICATIONS
DOI: 10.1002/adsc.200600356
Borate as a Phosphate Ester Mimic in Aldolase-Catalyzed Reac-
tions: Practical Synthesis of l-Fructose and l-Iminocyclitols
Masakazu Sugiyama,a Zhangyong Hong,a Lisa J. Whalen,a
William A. Greenberg,a,* and Chi-Huey Wonga,*
a
Department of Chemistry and the Skaggs Institute for Chemical Biology, The Scripps Research Institute, 10550 N. Torrey
Pines Rd., La Jolla, CA 92037, USA
Fax : (+1)-858-784-2409; e-mail: wong@scripps.edu or chwong@gate.sinica.edu.tw
Received: July 14, 2006; Accepted: November 7, 2006
Supporting information for this article is available on the WWW under http://asc.wiley-vch.de/home/.
Abstract: Dihydroxyacetone phosphate (DHAP)-
dependent aldolases have been widely used for the
organic synthesis of unnatural sugars or derivatives.
The practicality of using DHAP-dependent aldolas-
es is limited by their strict substrate specificity and
the high cost and instability of DHAP. Here we
report that the DHAP-dependent aldolase l-rham-
nulose 1-phosphate aldolase (RhaD) accepts dihy-
droxyacetone (DHA) as a donor substrate in the
presence of borate buffer, presumably by reversible
in situ formation of DHA borate ester. The reaction
appears to be irreversible, with the products ther-
modynamically trapped as borate complexes. We
have applied this discovery to develop a practical
one-step synthesis of the non-caloric sweetener l-
fructose. l-Fructose was synthesized from racemic
glyceraldehyde and DHA in the presence of RhaD
and borate in 92% yield on a gram scale. We also
synthesized a series of l-iminocyclitols, which are
potential glycosidase inhibitors, in only two steps.
complex with a DHAP analogue, showing that the
phosphate moiety forms hydrogen bonds with four
residues in the active site, and is the major determi-
nant for binding affinity between the aldolase and
DHAP.[2] This likely explains why unphosphorylated
analogues such as DHA are not accepted by these en-
zymes. Although several chemical[3] or chemo-enzy-
matic[4] syntheses of DHAP have been described, the
high cost and instability of DHAP, as well as the re-
quirement of a phosphatase to remove the phosphate
ester from the product, makes these aldolases less
than ideally practical.
Previous efforts to overcome the DHAP depend-
ence of aldolases have involved in situ formation of
arsenate or vanadate esters of DHA, which act as
phosphate ester mimics.[5–7] Although arsenate has
been used effectively for synthetic applications,[5,6] the
high toxicity of arsenate is an issue from a practical
standpoint. Vanadate can also be used even at lower
concentrations than arsenate, but vanadate may have
problematic redox activity and its expense limits its
practical utility.[5]
Keywords: aldol reaction; aldolases; borate; imino-
cyclitols; l-sugars; synthetic methods
Here we report that using inorganic borate buffer
allows DHA to be accepted as a substrate by RhaD
aldolase, presumably by reversibly forming a borate
ester with dihydroxyacetone (DHA) in situ. We have
used this discovery to develop a practical, inexpensive
Dihydroxyacetone phosphate (DHAP)-dependent al- one-step synthesis of l-fructose (Figure 1). The un-
dolases catalyze aldol reactions using DHAP as a nu- natural non-caloric sweetener l-fructose has been the
cleophilic donor substrate, and have been synthetical- target of several chemical[8] and chemoenzymatic syn-
ly useful due to their ability to form carbon-carbon theses,[9] and we recently reported a one-pot synthesis
bonds with high stereoselectivities and enantioselec- of l-fructose from DHAP and dl-glyceraldehyde
[10]
tivities.[1] A major drawback of DHAP-dependent al- using RhaDaldolase and acid phosphatase.
We de-
dolases is their strict donor substrate specificity scribe in this communication the development of
toward DHAP, and non-phosphorylated dihydroxy- RhaD-catalyzed one-step synthesis of l-fructose from
acetone (DHA) cannot be utilized by the DHAP-de- dl-glyceraldehyde and DHA in the presence of
pendent aldolases. The crystal structure of rhamnu- borate on a gram scale. We also developed two-step
lose 1-phosphate aldolase (RhaD), which is a class II syntheses of a series of l-iminocyclitols, which are
DHAP-dependent aldolase, has been reported in promising as glycosidase inhibitors.
Adv. Synth. Catal. 2006, 348, 2555 – 2559
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2555

Masakazu Sugiyama et al.
COMMUNICATIONS
required for increasing l-fructose yield. In the case of
the aldolase reaction with arsenate ester, d-fructose
1,6-diphosphate (FDP) aldolase also needed a 6-fold
excess of arsenate (120 mM) against 20 mM DHA.[6]
The excess borate may play a role in trapping the l-
fructose product and preventing the reverse reaction,
as described later in this communication. From a
practical standpoint, sodium borate is a common and
inexpensive buffer, and the one-step synthesis with
RhaDfrom DHA can be performed simply by replac-
ing traditionally used buffers with sodium borate.
Next, DHA concentrations were varied at constant
concentration of 100 mM dl-glyceraldehyde and
200 mM borate (Figure 2b). l-Fructose production
Figure 1. One-step synthesis of l-fructose and l-rhamnulose,
from dihydroxyacetone and dl-glyceraldehyde or lactalde-
hyde using RhaDaldolase in the presence of borate.
His-tagged RhaDaldolase was overexpressed in E. was increased as the initial DHA concentration in-
coli and purified as reported previously.[11] Alterna- creased, and reached maximum at 4 equivalents of
tively, E. coli BL21 (DE3) cells harboring pETDR- DHA relative to l-glyceraldehyde.
haDwere used directly as a whole cell biocatalyst
A gram-scale synthesis of l-fructose was performed
without enzyme purification. A typical reaction mix- under the optimized reaction conditions, resulting in
ture contained borate buffer (100 mM, pH 7.6), DHA an isolated yield of 92% l-fructose based on l-glycer-
(50 mM), dl-glyceraldehyde (100 mM), and RhaDal-
aldehyde (see Experimental Section). We used inex-
dolase. To confirm that this is an aldolase-catalyzed pensive dl-glyceraldehyde instead of enantiomerically
reaction, an omission test was performed under these pure l-glyceraldehyde as the acceptor substrate. If the
conditions. The results showed that all components enzyme were to accept d-glyceraldehyde as a sub-
(borate, DHA, glyceraldehyde, and RhaD) were nec- strate, the expected product would be d-sorbose.
essary for formation of l-fructose, indicating that this However, d-sorbose was not observed in the crude
is a borate-dependent RhaD-catalyzed aldol reaction. product NMR analysis. The unreacted d-glyceralde-
We believe that inorganic borate reversibly forms hyde was easily separated from l-fructose product by
esters with DHA in situ in aqueous solution, and thus chromatography.
formed DHAB can be accepted by RhaD aldolase as
A one-pot synthesis of l-rhamnulose, the unphos-
a donor substrate. After the aldol reaction with l- phorylated form of the natural substrate of RhaD,
glyceraldehyde to give l-fructose borate ester, the was also performed (Figure 1). In situ formation of
metastable borate esters were readily hydrolyzed dl-lactaldehyde by ozonolysis of commercially avail-
during reaction work-up.
able 2-hydroxy-3-butene was followed by aldol reac-
The effect of borate concentration on the aldol re- tion with DHA in the presence of borate and RhaD,
action with constant DHA (50 mM) and dl-glyceral- to afford l-rhamnulose in 53% yield after purifica-
dehyde (100 mM) was examined (Figure 2a). The re- tion.
action reached maximum yield at about 200 mM
Enzyme kinetics in the presence of borate were de-
borate. Although this reaction could conceivably be termined in both the forward and reverse direction,
catalytic in borate, excess equivalents of borate were and compared to the kinetics for the phosphorylated
Figure 2. l-Fructose synthesis from DHA and dl-GA. (a) Borate-dependent formation of l-fructose; sodium borate buffer
(10, 20, 50, 100, 200, 300, or 400 mM, pH 7.6), DHA (50 mM), dl-glyceraldehyde (100 mM). (b) DHA-dependent formation
of l-fructose; sodium borate buffer (200 mM, pH 7.6), DHA (10, 20, 50, 75, 100, 150, or 200 mM), and dl-glyceraldehyde
(100 mM).
2556
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 2555 – 2559

Practical Synthesis of l-Fructose and l-Iminocyclitols
COMMUNICATIONS
substrates (Table 1). The retroaldol activity of RhaD
for l-rhamnulose in the presence of 100 mM borate
was approximately 60 times lower than for the natural
substrate l-rhamnulose 1-phosphate. Moreover, the
Table 1. Activities of RhaDwith non-phosphorylated sub-
strates in the presence of borate.
Substrate
Vmax (mmol/
Km
(mM)
minmgÀ1)
l-rhamnulose 1-phosphate
l-fructose 1-phosphate
l-rhamnulose+borate
l-fructose+borate
2.2
0.71
0.035
not detected
0.96
3.2
dl-lactaldehyde+DHAP
dl-lactaldehyde+DHA+
borate
33
1.0
dl-glyceraldehyde+DHAP
dl-glyceraldehyde+DHA+
borate
23
0.48
retroaldol activity for l-fructose in the presence of
borate was undetectable. Increasing the incubation
period or the amount of enzyme did not yield the ret-
roaldol products (DHA and l-glyceraldehyde), as
checked photometrically, nor by HPLC analysis.
These results suggest that l-fructose borate esters are
Figure 3. Two-step synthesis of iminocyclitols 1–4 from dihy-
droxyacetone using RhaDaldolase in the presence of
borate, followed by reductive cyclization.
In conclusion, we have discovered that RhaDaldo-
not active substrates for the retroaldol reaction, while lase accepts DHA as a donor substrate in the pres-
DHA-borate can be efficiently accepted by RhaD to ence of borate buffer, presumably by reversible in situ
go in the synthetic direction. Literature precedent formation of DHA borate esters. Borate esters are
from 11B, 13C NMR studies[12] and thermodynamic formed spontaneously and reversibly in aqueous solu-
studies[13] suggest that d-fructose forms relatively tion, resulting in a simple one-step operation in which
stable 1:1 and 1:2 borate-b-d-fructofuranose com- the borate esters are formed in situ, and hydrolyzed
plexes. Thus, the l-fructose product may be thermo- during the work-up. This contrasts with the case of
dynamically trapped in such complexes and prevented phosphate esters, which have to be prepared and then
from undergoing the retroaldol reaction.
hydrolyzed in separate steps. In addition to reducing
Next we applied our method to the facile synthesis the synthesis to a single step, we have dramatically re-
of a series of l-iminocyclitols by using azido aldehyde duced the cost of producing l-fructose. Whereas
acceptors to efficiently synthesize several l-iminocy- DHAP is commercially available for ca. $ 2200/gram,
clitols (Figure 3). Following known procedures[14] the DHA costs only $ 0.20/gram. We also applied this
azido ketone aldol products underwent diasteroselec- method to develop facile two-step syntheses of l-imi-
tive reductive cyclization to produce the iminocycli- nocyclitols.
tols in only two steps. Recent reports have indicated
Borate is a known inhibitor of some enzymes, due to
that the l-enantiomers of known d-iminocyclitol gly- esterification of hydroxyl groups of enzyme residues,[19]
cosidase inhibitors can also be potent glycosidase in- substrates,[20] or by mimicking tetrahedral transition
hibitors, acting in a non-competitive mode and with states.[19,21] However, here we have demonstrated that
unique specificity profiles.[15] While compound 1 (l- borate can be used as a phosphate ester mimic for a
deoxymannojirimycin) is well-characterized, com- practical, inexpensive enzymatic synthesis on gram
pounds 2[16] and 3a[17] and 3b[15a] have not been exten- scale. Investigation of new applications of borate esters
sively studied for inhibitory activity, and 4 has not with other DHAP-dependent aldolases and other
been reported before. In the case of 3, RhaDprefer-
entially accepts the d-enantiomer of the aldehyde
over the l-enantiomer by a ratio of 2:1, perhaps due
to the different steric or hydrogen-bonding con-
straints. For 4, only the product from the d-aldehyde
was observed. Its structure was assigned by correla-
tion with the known enantiomer of 4.[18]
classes of phosphate-utilizing enzymes is ongoing.
Experimental Section
Synthetic and analytical procedures, enzyme expression, and
kinetics experiments are described in the Supporting Infor-
Adv. Synth. Catal. 2006, 348, 2555 – 2559
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
2557

Masakazu Sugiyama et al.
COMMUNICATIONS
mation. A typical gram-scale synthesis of l-fructose is de-
scribed below.
Chem. 2004, 69, 9310–9312; c) E. L. Ferroni, V. DiTel-
la, N. Ghanayem, R. Jeske, C. Jodlowski, M. OꢁCon-
nell, J. Styrsky, R. Svoboda, A. Venkataraman, B. M.
Winkler, J. Org. Chem. 1999, 64, 4943–4945; d) O.
Meyer, M. Rohmer, C. Grosdemange-Billiard, Tetrahe-
dron Lett. 2004, 45, 7921–7923.
RhaD-Catalyzed Synthesis of l-Fructose in the
Presence of Borate
To a solution of dl-glyceraldehyde (1.80 g, 20 mmol) and
DHA (3.60 g, 40 mmol) in water (160 mL), 1M sodium
borate buffer (40 mL, pH7.6) and toluene (400 mL) were
added, and E. coli BL21 (DE3) cells harboring pETDRhaD
(2.4 g by wet weight) were suspended. The reaction mixture
was shaken at 378C for 16 h, and cells were removed by cen-
trifuge. The reaction mixture was passed through a column
of Amberlite IR-120 (H+) resin (50 mL) and then eluted
with additional water. The resulting solution was passed
through a column of Amberlite IRA-743 resin (120 mL) to
remove borate.[22] After evaporation, the mixture was puri-
fied using silica gel chromatography with ethyl acetate/
methanol/water (40/10/7) as eluent. Fractions containing l-
fructose were collected and concentrated to afford the prod-
uct; yield: 1.66 g (9.2 mmol, 92% based on l-glyceralde-
[4] a) C.-H. Wong, G. M. Whitesides, J. Org. Chem. 1983,
48, 3199–3205; b) D. C. Crans, G. M. Whiteside, J. Am.
Chem. Soc. 1985, 107, 7019–7027; c) W.-D. Fessner, G.
Sinerius, Angew. Chem. Int. Ed. Engl. 1994, 33, 209–
212; d) W.-D. Fessner, C. Walter, Angew. Chem. Int.
Ed. Engl. 1992, 31, 614–616; e) F. Charmantray, P.
Dellis, S. Samreth, L. Hecquet, Tetrahedron Lett. 2006,
47, 3261–3263; f) T. van Herk, A. F. Hartog, H. E.
Schoemaker, R. Wever, J. Org. Chem. 2006, ASAP.
[5] D. G. Drueckhammer, J. R. Durrwachter, R. L. Peder-
son, D. C. Crans, L. Daniels, C.-H. Wong, J. Org. Chem.
1989, 54, 70–77.
[6] R. Schoevaart, F. van Rantwijk, R. A. Sheldon, J. Org.
Chem. 2001, 66, 4559–4562.
1
hyde). H NMR and 13C NMR spectra in D2O were identical
[7] a) R. Lagunas, A. Sols, FEBS Lett. 1968, 1, 32–34;
b) D. C. Crans, K. Sudhakar, T. J. Zamborelli, Biochem-
istry. 1992, 31, 6812–6821.
[8] a) T. Matsumoto, T. Enomoto, T. Kurosaki, J. Chem.
Soc., Chem. Commun. 1992, 8, 610; b) C. C. Cheng,
R. L. Whistler, Carbohydr. Res. 1988, 175, 265–271;
c) Y. Gizaw, J. N. BeMiller, Carbohydr. Res. 1995, 266,
81–85.
[9] a) J. W. Mayo, R. L. Anderson, Carbohydr. Res. 1968, 8,
344–347; b) M. R. Dhawale, W. A. Szarek, G. W. Hay,
A. M. Kropinski, Carbohydr. Res. 1986, 155, 262–265;
c) H. Itoh, K. Izumori, J. Ferment. Bioeng. 1996, 81,
351–353; d) I. Henderson, K. B. Sharpless, C.-H. Wong,
J. Am. Chem. Soc. 1994, 116, 558.
to those of authentic d-fructose spectra and l-fructose spec-
tra (see Supporting Information). NMR samples were al-
lowed to equilibrate for 1 hour in D2O to allow fructose to
reach its equilibrium mixture of furanose and pyranose
forms; [a]2D3: +93.18 (c 3, H2O).
Supporting information
For spectroscopic and analytical data, synthetic and analyti-
cal procedures, enzyme expression, and kinetics experi-
ments, see the supporting information.
[10] D. Franke, T. Machajewski, C.-C. Hue, C.-H. Wong, J.
Acknowledgements
Org. Chem. 2003, 68, 6828–6831.
[11] a) L. Vidal, O. Durany, T. Suau, P. Ferrer, M. D. Be-
naiges, G. Caminal, J. Chem. Technol. Biotechnol. 2003,
78, 1171–1179; b) E. Garcia-Junceda, G. J. Shen, T.
Sugai, C.-H. Wong, Bioorg. Med. Chem. 1995, 3, 945–
953.
We thank the N.I.H. for financial support (GM044154).
L. J. W. thanks the Skaggs Institute for Chemical Biology for
a postdoctoral fellowship. M.S. was supported by Ajinomoto
Co. Inc.
[12] a) S. Chapelle, J.-F. Verchere, Tetrahedron 1988, 44,
4469–4482; b) S. Chapelle, Carbohydr. Res. 1989, 191,
63–70; c) G. R. Kennedy, M. J. How, Carbohydr. Res.
1973, 28, 13–19; d) H. Pelmore, M. C. R. Symons, Car-
bohydr. Res. 1986, 115, 206–211.
References
[1] a) H. J. M. Gijsen, L. Qiao, W. Fitz, C.-H. Wong, Chem.
Rev. 1996, 96, 443–473; b) S. Takayama, G. J. McGar-
vey, C.-H. Wong, Annu. Rev. Microbiol. 1997, 51, 285–
310; c) W.-D. Fessner, J. Badia, O. Eyrisch, A. Schneid-
er, G. Sinerius, Tetrahedron Lett. 1992, 33, 5231–5234;
d) W.-D. Fessner, G. Sinerius, A. Schneider, M. Dreyer,
G. E. Schulz, J. Badia, J. Aguilar, Angew. Chem. Int.
Ed. Engl. 1991, 30, 555–558; e) R. Schoevaart, F. van
Rantwijk, R. A. Sheldon, J. Org. Chem. 2000, 65, 6940;
f) L. Espelt, J. Bujons, T. Parella, J. Calveras, J. Joglar,
A. Delgado, P. Clapes, Chem. Eur. J. 2005, 11, 1392–
1401.
[13] R. Aruga, J. Chem. Soc., Dalton Trans. 1988, 2971–
2974.
[14] S. Takayama, R. Martin, J. Wu, K. Laslo, G. Siuzdak,
C.-H. Wong, J. Am. Chem. Soc. 1997, 119, 8146–8151.
[15] a) C.-Y. Yu, N. Asano, K. Ikeda, M.-X. Wang, T. D.
Butters, M. R. Wormald, R. A. Dwek, A. L. Winters,
R. J. Nash, G. W. J. Fleet, Chem. Commun. 2004, 1936–
1937; b) N. Asano,
et al., J. Med. Chem. 2005, 48,
2036–2044; c) N. Asano, K. Ikeda L. Yu, A. Kato, K.
Takebayashi, I. Adachi, I. Kato, H. Ouchi, H. Takahata,
G. W. J. Fleet, Tetrahedron: Asymmetry 2005, 16, 223–
229.
[2] M. Kroemer, I. Merkel, G. E. Schulz, Biochemistry
2003, 42, 10560–10568.
[16] G. W. J. Fleet, S. J. Nicholas, P. W. Smith, S. V. Evans,
L. E. Fellows, R. J. Nash, Tetrahedron Lett. 1985, 26,
3127–3130.
[3] a) S. H. Jung, J.-H. Jeong, P. Miller, C.-H. Wong, J. Org.
Chem. 1994, 59, 7182; b) F. Charmantray, L. El Blidi, F.
Gefflaut, L. Hecquet, J. Bolte, M. Lemaire, J. Org.
2558
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 2555 – 2559

Practical Synthesis of l-Fructose and l-Iminocyclitols
COMMUNICATIONS
[17] R. E. Lee, M. D. Smith, R. J. Nash, R. C. Griffiths, M. [20] K. Wada, Y. Asano, Appl. Environ. Microbiol. 2003, 69,
McNeil, R. K. Grewal, W. Yan, G. S. Besra, R. J.
Brennan, G. W. J. Fleet, Tetrahedron Lett. 1997, 38,
6733–6736.
7058–7062.
[21] a) D. F. Iwig, A. Uchida, J. A. Stromberg, S. J. Booker,
J. Am. Chem. Soc. 2005, 127, 11612–11613; b) R.
Baggio, D. Elbaum, Z. F. Kanyo, P. J. Carroll, R. C.
Cavalli, D. E. Ash, D. W. Christianson, J. Am. Chem.
Soc. 1997, 119, 8107–8108.
[18] Y. Takaoka, T. Kajimoto, C.-H. Wong, J. Org. Chem.
1993, 58, 4809–4812.
[19] a) M. Wolny, Eur. J. Biochem. 1977, 80, 551–556;
b) S. S. Tate, A. Meister, Proc. Natl. Acad. Sci. USA
1978, 75, 4806–4809; c) T. R. Transue, J. M. Krahn,
S. A. Gabel, E. F. DeRose, R. E. London, Biochemistry
2004, 43, 2829–2838.
[22] K. B. Hicks, G. L. Simpson, A. G. W. Bradbury, Carbo-
hydr. Res. 1986, 147, 39–48.
Adv. Synth. Catal. 2006, 348, 2555 – 2559
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
2559